Volume 63, Issue 3, Pages 397-405 (August 2009) Category-Specific Organization in the Human Brain Does Not Require Visual Experience  Bradford Z. Mahon, Stefano Anzellotti, Jens Schwarzbach, Massimiliano Zampini, Alfonso Caramazza  Neuron  Volume 63, Issue 3, Pages 397-405 (August 2009) DOI: 10.1016/j.neuron.2009.07.012 Copyright © 2009 Elsevier Inc. Terms and Conditions

Figure 1 Main Effects of Task in Sighted and Congenitally Blind Participants The ventral object processing stream was functionally defined as all voxels that were significant in the omnibus test (random effects analysis) in sighted participants viewing pictures and that were at or below a z coordinate in Talairach space of 6 (resulting mask = 131,602 mm3). The contrast map for the omnibus test for viewing pictures within the functionally defined ventral stream is shown in the right panel. The left and middle panels show the omnibus analysis when sighted (left) and congenitally blind (middle) participants performed the auditory size-judgment task (collapsing across animal and nonliving stimuli). As can be seen, for the auditory size-judgment task there was a pattern of relatively decreased BOLD responses in early visual regions in sighted participants but increased BOLD responses in congenitally blind participants. Neuron 2009 63, 397-405DOI: (10.1016/j.neuron.2009.07.012) Copyright © 2009 Elsevier Inc. Terms and Conditions

Figure 2 Regions of Interest Analyses of Category Preferences Regions of interest (ROIs) were defined contrasting “animal” against “nonliving” stimuli in sighted participants viewing pictures (thresholded at p < .05, FDR corrected). Voxels showing differential BOLD responses for animal stimuli compared to nonliving stimuli are shown on the red-yellow color scale (panel A), while voxels showing differential BOLD responses for nonliving things compared to animals are shown on the blue-green color scale (panel B). The bar graphs depict the estimates for BOLD responses for animal and nonliving stimuli within those ROIs, for all datasets (collapsing across all voxels within the ROI). The left-most graphs (panels A and B) showing BOLD responses for sighted participants performing auditory size-judgments do not have indicators for statistical significance because those data come from voxels that were used to define the ROI. Error bars reflect the standard error of the mean. Neuron 2009 63, 397-405DOI: (10.1016/j.neuron.2009.07.012) Copyright © 2009 Elsevier Inc. Terms and Conditions

Figure 3 Contrast Maps for Animal versus Nonliving Stimuli for All Groups of Participants Voxels showing differential BOLD responses for animal stimuli compared to nonliving stimuli are shown on the red-yellow color scale (panel A), while voxels showing differential BOLD responses for nonliving things compared to animals are shown on the blue-green color scale (panel B). For visualization purposes, all statistical contrast maps were thresholded at p < .01, uncorrected. Within the left lateral occipital ROI (see Figure 2), the voxels showing the greatest difference between animal and nonliving stimuli were as follows: sighted participants auditory size-judgments: −43, −76, −5; peak effect: t = 3.61, p < .001; congenitally blind: −43, −76, −7; peak effect: t = 2.49, p < .02; sighted participants viewing pictures: −42, −76, −2; peak effect: t = 5.44, p < .001. Within the bilateral medial ventral stream ROIs (see Figure 2), the voxels showing the greatest difference between animal and nonliving stimuli were as follows: sighted participants auditory size-judgment: left: −24, −40, −11; peak effect: t = −9.20, p < .001; right: 27, −31, −17; t = −7.96, p < .001; congenitally blind: left: −30, −46, −8; peak effect: t = −9.25, p < .001; right: 30, −37, −11; peak effect: t = −6.21, p < .001; sighted participants viewing pictures: left: −24, −40, −11; peak effect: t = −6.88, p < .001; right: 27, −49, −8; peak effect: t = −13.21, p < .001. Neuron 2009 63, 397-405DOI: (10.1016/j.neuron.2009.07.012) Copyright © 2009 Elsevier Inc. Terms and Conditions

Figure 4 Overlay of Contrast Maps of Category Preferences in Sighted and Blind Participants All statistical contrast maps (nonliving versus living) were thresholded at p < .01, uncorrected (as in Figure 3). (A) Regions showing category preferences are plotted for sighted participants viewing pictures and sighted participants performing the auditory size-judgment task. The greatest overlap is observed for nonliving things in medial regions on the ventral surface of occipital-temporal cortex. Voxels in lateral occipital cortex showing differential BOLD responses for living things in sighted participants performing auditory size-judgments fall within the set of voxels showing the same category preference for the picture viewing experiment. (B) Regions showing category preferences are plotted for sighted participants viewing pictures and congenitally blind participants performing the auditory size-judgment task. As was observed for sighted participants performing the auditory size-judgment task (panel A), the regions showing differential BOLD responses for nonliving things were larger than those showing differential BOLD responses for living things. Voxels in blind participants in lateral occipital cortex showing a category preference for living things fall within the region identified in sighted participants viewing pictures. Neuron 2009 63, 397-405DOI: (10.1016/j.neuron.2009.07.012) Copyright © 2009 Elsevier Inc. Terms and Conditions

Figure 5 Medial-to-Lateral Index Computed over Ventral Occipital-Temporal Cortex (A) The color overlay shows the extent of the ROI in ventral occipital-temporal cortex (maximum “y” Talairach = −20, minimum “y” Talairach = −70; following Pietrini et al. (2004). Within the ventral occipital-temporal ROI, a medial-to-lateral index was calculated, by averaging the contrast-weighted t values (animal stimuli – nonliving stimuli) along the anterior-posterior (y) and superior-inferior (z) dimensions, within the range of |25| to |40| on the x axis in Talairach space. The results of this analysis are plotted in panels (B) for sighted participants viewing pictures, (C) for sighted participants performing auditory size-judgments, and (D) for blind participants. Error bars on all graphs represent the standard error of the mean for contrast-weighted t values, averaged along the z and y axes. Vertical red dotted lines are placed at the mean t value, in order to indicate the corresponding Talairach coordinate. Neuron 2009 63, 397-405DOI: (10.1016/j.neuron.2009.07.012) Copyright © 2009 Elsevier Inc. Terms and Conditions